Stem CO 2 Efﬂux as an Indicator of Forests’ Productivity in Relict Juniper Woodlands ( Juniperus thurifera L.) of Southern Spain

: There are considerable uncertainties about the C cycle in semi-arid ecosystems. Hence, studies that have focused on Juniperus in Mediterranean woodlands are non-existent. This study provides a survey of the effect of the juniper woodland type (young and mature woodlands; joint effect of maturity and forest productivity) on stem respiration. We checked the seasonal variation of stem respiration, evaluating the effects of stem temperature on stem CO 2 efﬂux. For this, we measured the stem CO 2 efﬂux ( µ mol CO 2 m − 2 s − 1 ) over the four seasons on 16 junipers using LI-6400 equipment. The results showed that in the more productive site (young woodland), the stem CO 2 efﬂux was higher. This variable followed a clear seasonal trend, being higher during the spring and progressively decreasing in cold periods. In both juniper woodlands, and especially in the older forests, the Q 10 coefﬁcients were low (<2), typical of cold forests and slow-growing species. The exponential model also conﬁrmed that the Q 10 was signiﬁcantly higher in young juniper trees. Thus, stem CO 2 efﬂux was an indicator of the growth in this juniper woodland that is well adapted to a semi-arid climate.


Introduction
Autotrophic respiration is an important component of ecosystem-scale carbon budgets. In forests, Net Primary Production (NPP) is defined as the difference between gross primary production and autotrophic respiration [1]. Soil autotrophic respiration is the CO 2 produced from plant root and microbials metabolism, whereas aboveground autotrophic respiration has two major components: total wood respiration, which comprises stem and branch respiration, and foliar respiration [2]. In this balance, stem respiration makes an important contribution to total ecosystem respiration. It was estimated that the annual CO 2 efflux from stem tissues consumed 26% of the annual gross primary production [3]. Therefore, stem respiration is regarded as an important factor in the regulation of forest productivity and C storage, and it is one of the main components of the CO 2 efflux that release C to the atmosphere in forest ecosystems [4]. Although it does not usually reach the importance (in magnitude) of soil respiration, it is a variable component of the C cycle that allows us to analyze forests from the point of view of their productivity [2]. Thermophilic juniper woodlands occupy the driest areas of the distribution of this species in Spain [21], and they constitute relict forests with slow-grow [24] that can be classified into two levels of maturity and productivity [25]. The first are old or mature woodlands, and these generally contain trees that are over 150 years old and are characterized by stony soils with a low tree density. The second level is young woodlands, dominated by trees that are less than 100 years old. Young woodlands grow in abandoned farmlands (i.e., areas with better soils), and thus these forests have higher growth rates and tree density, presenting greater productivity [25,26].
The characteristics of the two thermophilic Spanish juniper woodlands that exist allowed us to study the stem CO 2 efflux in two forests with different degrees of maturity and productivity (due to age and soil quality) in a semi-arid Mediterranean climate and throughout the different seasons. We can hypothesize that variation in CO 2 efflux between stems is proportional to differences in the respiratory activity from these woody tissues. For this, seasonal estimations of stem respiration are required under naturally changing conditions of growth [7].
This main hypothesis of our research is applied on the fact that the type of juniper woodland, growing season (seasonality), and different climatic conditions (which are mainly reflected in the stem temperature) should significantly affect the CO 2 efflux of the juniper stems. Therefore, the CO 2 efflux measured from the stem and that which mainly comes from stem respiration should be an estimator of the productivity of the two juniper woodlands.
Thus, the specific objectives of the study were the following: (i) to analyze the effects of the juniper woodland type (young and mature woodlands; effect of soil quality and forests productivity) on the stem CO 2 efflux; (ii) to check the seasonal variation of the stem CO 2 efflux in the two juniper woodlands; (iii) to evaluate the effects of stem temperature on stem CO 2 efflux for both types of woodlands, which have different growth rates.

Study Area
The study was carried out in the Spanish juniper woodlands of El Campo de Montiel Mountains (southern Spain; Figure 1). These forests must be classified as thermophilic according to [21]. The climate of the area is semi-arid cold (BSk type, KÖPPEN [27]), continental, with extreme temperatures between 43 and −21 • C (data from "El Bonillo" station: 38 • 57 N, 01 • 09 W, 1068 m.a.s.l., period of 30 years; this station is located within 7 km of the research area, and owned by the State Meteorological Agency of Spain), with an average annual temperature of 12.8 • C. The precipitations correspond to a dry ombroclimate, with annual rainfall of 450 mm.
The juniper trees in which stem CO 2 efflux was measured were selected from the two types of thermophilic juniper woodlands ( Figure 2 and Table 1): (i) a mature woodland with an average age of 170 years, growing in lithic leptosols (soil depth less than 15 cm) with stands of moderate slope (0-10%) and (ii) a young woodland, with a mean age of 66 years, which grows in abandoned farmlands and with deeper soils.
In each type of juniper woodland, a homogenous experimental area (approximately 1 hectare) was selected, representative of the forests studied. Both areas are less than 1 km apart, ensuring that both stands are growing under the same climatic conditions. Thus, both experimental areas were similar in elevation (1050 m.a.s.l.), slope (gentle slopes; average 5 • ), exposure (paramo formation with flat surfaces), and total rainfall (450 mm) due to their close proximity.  [21]. UTM coordinates (ETRS89).
The juniper trees in which stem CO2 efflux was measured were selected from the two types of thermophilic juniper woodlands ( Figure 2 and Table 1): (i) a mature woodland with an average age of 170 years, growing in lithic leptosols (soil depth less than 15 cm) with stands of moderate slope (0-10%) and (ii) a young woodland, with a mean age of 66 years, which grows in abandoned farmlands and with deeper soils.   [21]. UTM coordinates (ETRS89).  [21]. UTM coordinates (ETRS89).
The juniper trees in which stem CO2 efflux was measured were selected from the two types of thermophilic juniper woodlands ( Figure 2 and Table 1): (i) a mature woodland with an average age of 170 years, growing in lithic leptosols (soil depth less than 15 cm) with stands of moderate slope (0-10%) and (ii) a young woodland, with a mean age of 66 years, which grows in abandoned farmlands and with deeper soils.   The two juniper woodlands studied are located in lime-rich substrates (Jurassic dolomites), thus presenting a basic pH ( Table 1). The soils in the young woodland have a greater clay content than those in mature woodlands. Conversely, the mature woodland soils contain more sand. Thus, bulk density was slightly greater in the mature woodland. The existence of a greater amount of organic C in the mature woodland is noteworthy, a consequence of the greater cumulative deposition of organic matter over the years. Biomass stocks and productivity were scarce in both woodlands, but the aboveground biomass in the young stand was similar to that of a more productive, paramo community [25]. We found that for mature woodlands, the few references that did exist for the Juniperus genus indicate a significantly larger biomass, and in the majority of studies carried out on other conifer species, biomass very often exceeded 50 Mg ha −1 [25]. Juniper trees in young woodland have a slow growth, and trees in mature woodland represent very slow-growing species.

Measurement of Stem CO 2 Efflux and Stem Temperature in Juniper Trees, and Soil Water Content
We measured the stem CO 2 efflux per unit of stem surface area (in µmol CO 2 m −2 s −1 ), assuming that the stem CO 2 efflux is a practical proxy for stem respiration [5]. Measurements were carried out in situ for 18 days, spread over the four seasons (3 days in winter, summer, and autumn and 9 days in spring, the more important growing season). Within the experimental areas, metallic collars (10.5 cm in diameter, and 5 cm in height) were installed (see methodology in Xu et al. [29]) on the stem of 16 adult junipers (8 per juniper woodland; to eliminate the effect of age) at a standardized height (breast height, 1.30 m above soil). Loose bark was removed and 100% silicone sealant was used to attach and seal the collars to the south-facing side of stem surfaces ( Figure 3). To check the variation of the respiratory flux along the stem, measurements were previously made at different heights of the trunk, although the height was not a factor that affected the CO 2 efflux (analysis not shown).
of the respiratory flux along the stem, measurements were previously made at different heights of the trunk, although the height was not a factor that affected the CO2 efflux (analysis not shown). The mean diameters at breast height of the junipers measured were 29.5 ± 12.4 cm, and 22.47 ± 9.8 cm for mature and young woodland, respectively (means ± standard deviation). Stem respiration was measured using the LI-6400 portable CO2 measurement equipment (LI-COR, Inc., Lincoln, NE, USA) to which the LI-6400-09 measurement chamber was attached. In each measurement, 3 repetitions of the CO2 efflux were taken, the mean of the measurements being the values for the data analysis. Measurements were made by reducing CO2 concentration inside the chamber and then letting it increase to an upper concentration limit (gas analyzer in closed system configuration). These limits changed depending on the respiration rate of the measured stem and the atmospheric CO2 concentration. For this, before starting each measurement, ambient CO2 concentration was measured and used as the target. This air CO2 concentration ranged between 372-411 p.p.m. along the full experimental period. The air flow (µmol air s −1 ) to pumping down was fixed depending on efflux level (200-700, for low and high efflux levels, respectively). The rate of change of CO2 inside the chamber was calculated in the three consecutive measurement cycles. Measurements were made at different times of the day to obtain greater variability in the dataset. The total dataset in the analysis included 288 measures.
The stem temperature (Ts; °C) was measured using temperature sensors installed next to the cylinders and with the sensor in contact with the sapwood. Stem temperature was recorded on a half-hourly basis close to each collar using a temperature probe (PB-5002-1M5, Gemini data loggers Ltd., Chichester, UK) inserted to a depth of 20 mm via a drilled (Ø < 5 mm) hole sealed with silicone; data were monitored with Tinytag Ultra 2 TGU-4020 data loggers (Gemini data loggers Ltd., Chichester, UK). Temperature data were automatically stored in the data loggers and associated with the time of efflux measurements.
Finally, soil water content (SWC, %) was measured with a moisture sensor (VH400-LV, Vegetronix, UT, USA), close to the sample tree. Measurements were recorded with HOBO data loggers (Onset Computers, Bourne, MA, USA). The moisture sensors were calibrated by the gravimetric method. Data loggers were downloaded while taking CO2 efflux measurements.

Data Analysis
To detect the effects of the woodland type and seasonality on stem CO2 efflux, GLM (multifactorial analysis of variance, MANOVA, p < 0.05) was performed including two fixed factors, juniper woodland type (W; two levels: young and mature woodland) and The mean diameters at breast height of the junipers measured were 29.5 ± 12.4 cm, and 22.47 ± 9.8 cm for mature and young woodland, respectively (means ± standard deviation). Stem respiration was measured using the LI-6400 portable CO 2 measurement equipment (LI-COR, Inc., Lincoln, NE, USA) to which the LI-6400-09 measurement chamber was attached. In each measurement, 3 repetitions of the CO 2 efflux were taken, the mean of the measurements being the values for the data analysis. Measurements were made by reducing CO 2 concentration inside the chamber and then letting it increase to an upper concentration limit (gas analyzer in closed system configuration). These limits changed depending on the respiration rate of the measured stem and the atmospheric CO 2 concentration. For this, before starting each measurement, ambient CO 2 concentration was measured and used as the target. This air CO 2 concentration ranged between 372-411 p.p.m. along the full experimental period. The air flow (µmol air s −1 ) to pumping down was fixed depending on efflux level (200-700, for low and high efflux levels, respectively). The rate of change of CO 2 inside the chamber was calculated in the three consecutive measurement cycles. Measurements were made at different times of the day to obtain greater variability in the dataset. The total dataset in the analysis included 288 measures.
The stem temperature (Ts; • C) was measured using temperature sensors installed next to the cylinders and with the sensor in contact with the sapwood. Stem temperature was recorded on a half-hourly basis close to each collar using a temperature probe (PB-5002-1M5, Gemini data loggers Ltd., Chichester, UK) inserted to a depth of 20 mm via a drilled (Ø < 5 mm) hole sealed with silicone; data were monitored with Tinytag Ultra 2 TGU-4020 data loggers (Gemini data loggers Ltd., Chichester, UK). Temperature data were automatically stored in the data loggers and associated with the time of efflux measurements.
Finally, soil water content (SWC, %) was measured with a moisture sensor (VH400-LV, Vegetronix, UT, USA), close to the sample tree. Measurements were recorded with HOBO data loggers (Onset Computers, Bourne, MA, USA). The moisture sensors were calibrated by the gravimetric method. Data loggers were downloaded while taking CO 2 efflux measurements.

Data Analysis
To detect the effects of the woodland type and seasonality on stem CO 2 efflux, GLM (multifactorial analysis of variance, MANOVA, p < 0.05) was performed including two fixed factors, juniper woodland type (W; two levels: young and mature woodland) and season (S; 4 levels: winter, spring, summer, and autumn) and their interaction (cross effect W × S), with the response variable being stem CO 2 efflux (Es; µmol CO 2 m −2 s −1 ): In the model 1, "J" is the juniper sample (16 levels) and J(W) is the nested effect of this juniper sample (16 junipers or "plots") within each woodland type (W). "Juniper" (J) was nested within "woodland" (W) as different juniper trees were measured in each woodland, and "J" we also considered a random factor as the sample trees were a random sample from the total junipers (randomized design, [30]). Thus, Model 1 represents a linear mixed-effects model with fixed and nested random effects [31]. In addition, µ is the overall mean of Es (µmol CO 2 m −2 s −1 ) and ε is the random residual. The rest of the interactions were excluded from the full model for their minimum significance and were included in the error term. The effect "between woodlands" on the response variable "within each season" was also analyzed (p < 0.05). Fisher's LSD test (Least Significant Difference) was used to make comparisons between means if the main factor was significant. It is remarkable that "woodland type"(W) is a confounded effect of both maturity (age) and soil quality, i.e., it represents the site productivity of the juniper woodlands.
To evaluate the effect of stem temperature (Ts; • C) and the juniper woodland type on stem CO 2 efflux (Es; µmol CO 2 m −2 s −1 ) throughout the seasons, the following model with a qualitative or dummy variable (woodland type: W = 0 for young woodland, and W = 1 for mature woodland) was fitted through multiple regression (Model 2): This model was fitted including all the variations in stem temperature measured in the field (n = 288). Model 2 shows an exponential trend that reflects the variation of the respiration of living plant cells and temperature, demonstrated in previous studies [5]. In our experiment, it was necessary to include the dummy variable (W; woodland type) to provide the joint effect of age and productivity (both effects are confounded), thus representing the two juniper stands. The selection of significant parameters in Model 2 was carried out by means of logarithmic transformation (Model 3; linearization of the exponential function; [32]) and stepwise regression ("forward selection" procedure): This corresponds with the comparison of two regression lines (Model 4 and 5): For young woodland: log(Es) = β + α × Ts + ε For mature woodland: The coefficients with a value of p < 0.05 were considered significant [31]. The significance of the model and the goodness of fit were assessed according to the F ratio (p < 0.05), the adjusted R 2 , and the standard error of the estimate (SEE). The Conditional Sums of Squares analysis was additionally used to determine whether the intercepts and the slopes of the two equations (Models 4 and 5) were significantly different. Observations with DIF > 2× n p were considered influential points (p is the number of coefficients, and n is the number of data [33]) and eliminated. Statistical analyses were performed using the Statgraphics Centurion XVIII ® software (Statgraphics Technologies, Inc., Virginia, USA).
Finally, in the significant equations of Model 3 (Model 4 and 5), the temperature coefficient Q 10 (rate of variation of Es when the stem temperature (Ts) increased by 10 • C [34]) was calculated. Thus, the significant equations were formulated in their exponential form (Model 6) [35,36]: Ts−Tr 10 Where Es is stem CO 2 efflux (µmol CO 2 m −2 s −1 ) at stem temperature Ts ( • C); Rr is the respiration at the reference temperature (Tr); and Q 10 is the temperature coefficient of respiration. In our study, the temperature of reference (Tr) was fixed at 0 • C, representing respiration in the non-growing or dormancy period [7,36].

Effects of Juniper Woodland Type on Stem CO 2 Efflux
The "woodland type" factor (W) significantly affected (F = 16.9; p < 0.00) the stem CO 2 efflux (Es), with this dependent variable being higher in the young juniper woodland (Tables 2 and 3). Table 2. MANOVA carried out (Model 1) to evaluate differences between the stem CO 2 efflux (Es, µmol CO 2 m −2 s −1 ) in the two juniper woodlands and throughout the four seasons (n = 288). The F ratios and the p-values are represented for the two main effects (woodland, season) and their interaction, as well as the effect of the woodland type within each season. The effects are significant if p < 0.05 (95% probability). LSD: least significant difference. W: woodland type; J: juniper tree (sample tree); S: season.

Effects
Total  Table 3. Mean values (± standard error) of stem CO 2 efflux (Es) in function of woodland type (W) and seasons (S). Averages of the levels within each effect followed by different letters reflect significant differences in the MANOVA performed (Fisher's LSD test, 95% probability, p < 0.05; n = 288). In addition, stem CO 2 efflux depending on the juniper woodland was found to be significant in the spring and summer (p < 0.00; Table 2), which seems to determine the more important growth period. In the winter and autumn, this significance was not found.

Seasonal Variation of Stem CO 2 Efflux in the Juniper Woodlands
The "season" factor (S) showed a significant effect (F = 28.6; p < 0.00) on the response variable Es (Table 2). Overall, the measurements of stem CO 2 efflux were higher in spring (mean value of 2.33 ± 0.09 µmol CO 2 m −2 s −1 ; Table 3). On the contrary, the stem CO 2 efflux was significantly reduced in winter (0.84 ± 0.15 µmol CO 2 m −2 s −1 ).
The interaction between the treatment variables was also significant (F = 3.32; p = 0.02; Table 2). Thus, significant differences were observed in stem CO 2 efflux depending on the woodland type (W). The measured CO 2 of juniper trees was higher in spring and summer for the young woodland (Figure 4), whereas young woodland maintained similar stem CO 2 efflux in these two seasons, and the stem respiration decreased from spring to summer in the older juniper woodland. In the cold seasons, winter and autumn, there were no significant differences between woodland type. CO2 efflux in these two seasons, and the stem respiration decreased from spring to summer in the older juniper woodland. In the cold seasons, winter and autumn, there were no significant differences between woodland type.
Soil water content (SWC; %) ranged from 24.1% (winter in young woodland) to 2.8% (summer in mature woodland; Figure 5). Seasonal soil moisture contents (mean ± standard error) were: winter (23.7 ± 2.5%), spring (14.1 ± 1.1%), summer (4.9 ± 2.1%), and autumn (11.4 ± 2.0%). Contrary to what happened with the trunk temperature, the seasonal variation of soil moisture was not correlated to stem CO2 efflux: when soil moisture was maximum, the efflux was minimal (winter), and the low value of soil moisture in summer did not correspond to the lower stem CO2 efflux rates.
No differences were found in stem temperature (°C) or soil water content (%) between the juniper sites in the same season (p < 0.05; Figure 5). Thus, climatic conditions were similar in both juniper woodlands at the time of efflux measurements.

Climatic Conditions: Stem Temperature (Ts; • C) and Soil Water Content (SWC; %)
Stem temperatures ranged from −2.7 • C to 27.9 • C (total range of 30.6 • C), and the mean stem temperature was 14.02 ± 8.88 • C (mean ± standard error). Seasonal stem temperatures (mean ± standard error) were: winter (3.34 ± 0.79 • C), spring (16.05 ± 0.46 • C), summer (25.12 ± 0.79 • C), and autumn (7.81 ± 0.78 • C). This seasonal behavior of stem temperature indicated a positive correlation between this variable and the stem CO 2 efflux: the higher stem CO 2 efflux were obtained under the higher values of stem respiration, i.e., in spring and summer ( Figure 5 and Table 3).

Effects of Stem Temperature on Stem CO2 Efflux
When adjusting the regression model of stem CO2 Efflux (Es) as a function of stem temperature (Ts; °C) and the juniper woodland type (Model 2), it was found that the dummy variable (W) was significant in the model, thus generating two independent equations for each juniper site (Table 4). This model was significant (F = 192, p < 0.00), and the Soil water content (SWC; %) ranged from 24.1% (winter in young woodland) to 2.8% (summer in mature woodland; Figure 5). Seasonal soil moisture contents (mean ± standard error) were: winter (23.7 ± 2.5%), spring (14.1 ± 1.1%), summer (4.9 ± 2.1%), and autumn (11.4 ± 2.0%). Contrary to what happened with the trunk temperature, the seasonal variation of soil moisture was not correlated to stem CO 2 efflux: when soil moisture was maximum, the efflux was minimal (winter), and the low value of soil moisture in summer did not correspond to the lower stem CO 2 efflux rates.
No differences were found in stem temperature ( • C) or soil water content (%) between the juniper sites in the same season (p < 0.05; Figure 5). Thus, climatic conditions were similar in both juniper woodlands at the time of efflux measurements.

Effects of Stem Temperature on Stem CO 2 Efflux
When adjusting the regression model of stem CO 2 Efflux (Es) as a function of stem temperature (Ts; • C) and the juniper woodland type (Model 2), it was found that the dummy variable (W) was significant in the model, thus generating two independent equations for each juniper site (Table 4). This model was significant (F = 192, p < 0.00), and the goodness of fit statistics reflected that the variability of Es was explained by Ts in a percentage of 71.4% (adjusted R 2 ), obtaining an SEE = 0.36 (relatively low). Table 4. Adjusted coefficients, standard errors, and statistics of the coefficients (t and p values) in Table 2. efflux (Es) as a function of stem temperature (Ts) and woodland type (W). In Model 2, the dependent variable was log-transformed-log(Es)-to obtain the intercept and slopes in function of woodland type (W). The summary statistics for the fitted model were adjusted R-squared (R 2 ) = 71.4%; standard error of estimation (SEE) = 0.36; mean absolute error = 0.30.  Table 5 shows the Conditional Sums of Squares analysis (ANOVA) to confirm whether the intercepts and the slopes of the two lines of Model 2 were significantly different. As the p-value was less than 0.05, the intercepts and slopes of the lines varied significantly amongst the two levels of the categorical factor "woodland type" (young and juniper woodland; 5% significance level). This corresponds to two separate regressions, and Model 3 can be written separately for each juniper woodland as follows:

Parameters
Young woodland (Model 4): log(Es) = −0.350 + 0.0684 × Ts + ε Mature woodland (Model 5): log(Es) = −0.480 + 0.057 × Ts + ε The generation of two significant equations from Model 2 confirmed different rates of stem CO 2 efflux for both types of juniper woodland as a function of the temperature. Thus, the exponential equation form (Model 6) was used to establish the relationship between the stem CO 2 efflux and stem temperature: Young woodland: Es = 0.704 × 1.98 Ts 10 Mature woodland: Es = 0.619 × 1.77 Ts 10 The exponential equations reflected that for a certain increase in stem temperature (Ts), the variation in stem CO 2 efflux (Es) was significantly greater in the young juniper woodland, which was the most productive ( Figure 6). Consequently, the temperature coefficient Q 10 (rate of respiration increase when the temperature increases by 10 • C) was significantly higher for the young woodland (1.98) than for the mature one (1.77). The respiration at 0 • C was 0.70 and 0.62 µmol CO 2 m −2 s −1 for young and mature woodland, respectively (significantly higher in the young woodland). We considered that at the temperature reference of 0 • C (R 0 ), diameter growth was negligible, and thus the respiration in this non-growing instance should be an estimator of the maintenance component. Regardless of the temperature of reference, Q 10 was invariant when the annual data were fitted by utilizing the exponential form.

Effects of Juniper Woodland Type on Stem CO2 Efflux
In the more productive site (young woodland), the annual rate of stem CO2 efflux was higher, and there would therefore be a greater respiratory activity. Parameters that define the productivity of the site, such as LAI, and % of sapwood in the stems, are correlated with greater stem respiration [1,14], a consequence of photosynthesis partially regulating stem growth and respiration [7]. The respiratory rate determines the amount of carbohydrates that can be invested in the maintenance (maintenance respiration) and development (growth respiration) of juniper trees, and therefore has an important relation on production of biomass [8]. Thus, measured CO2 efflux in juniper trees is the result of growth and maintenance stem respiration. There are an important link between total respiration and growth, even though growth respiration is only a percentage of total respiration (50% for wood [37]); maintenance respiration are also linked with the metabolic processes that will promote growth [38].
Additional factors could affect the correlation between stem respiration and measurements of stem CO2 efflux. Photosynthesis in bark or chlorenchyma can refix CO2 and hence lower net respiration rates for stems [3,4]. As the periderm blocks light to the photosynthetic tissue, photosynthetic rates are greatest in young stems and branches, and the effect of bark photosynthesis on net CO2 efflux from stems declines with stem size and age [12]; thus bark photosynthesis would be negligible in juniper trees with more than 50 years old. By carefully peeling off layers of the dead outer bark of juniper stems, and analyzing a cross section of trunk, the total absence of a green color indicated the absence of chlorophyll-containing tissues [39].

Effects of Juniper Woodland Type on Stem CO 2 Efflux
In the more productive site (young woodland), the annual rate of stem CO 2 efflux was higher, and there would therefore be a greater respiratory activity. Parameters that define the productivity of the site, such as LAI, and % of sapwood in the stems, are correlated with greater stem respiration [1,14], a consequence of photosynthesis partially regulating stem growth and respiration [7]. The respiratory rate determines the amount of carbohydrates that can be invested in the maintenance (maintenance respiration) and development (growth respiration) of juniper trees, and therefore has an important relation on production of biomass [8]. Thus, measured CO 2 efflux in juniper trees is the result of growth and maintenance stem respiration. There are an important link between total respiration and growth, even though growth respiration is only a percentage of total respiration (50% for wood [37]); maintenance respiration are also linked with the metabolic processes that will promote growth [38].
Additional factors could affect the correlation between stem respiration and measurements of stem CO 2 efflux. Photosynthesis in bark or chlorenchyma can refix CO 2 and hence lower net respiration rates for stems [3,4]. As the periderm blocks light to the photosynthetic tissue, photosynthetic rates are greatest in young stems and branches, and the effect of bark photosynthesis on net CO 2 efflux from stems declines with stem size and age [12]; thus bark photosynthesis would be negligible in juniper trees with more than 50 years old. By carefully peeling off layers of the dead outer bark of juniper stems, and analyzing a cross section of trunk, the total absence of a green color indicated the absence of chlorophyll-containing tissues [39].
Although the stem CO 2 efflux rates measured at both juniper woodlands fit the ranges previously reported (0.5-5 µmol CO 2 m −2 s −1 ) for forest species [6,40], the annual mean values of CO 2 efflux seem to be low (1.95 and 1.26 µmol CO 2 m −2 s −1 , for young and mature junipers, respectively), if we compare them with other conifer species. For example, [6] offers mean values (µmol CO 2 m −2 s −1 ) of 0.5-3.7 for Pinus canariensis, [34] 4-7 for Pinus ponderosa, and [41] 0.8-9 for Larix. This is an indicator that Spanish juniper trees behaves as a slow-growing species. The semi-arid climate with a warm and dry season alternating with a very cold, wet season is decisive for this behavior.

Seasonal Variation of Stem CO 2 Efflux in the Two Juniper Woodlands
Our results also reflect the existence of great seasonality in stem CO 2 efflux in both juniper woodlands, and it was higher during the growing season (spring) and progressively decreased from the warm to cold periods. Climatic seasonality in Mediterranean ecosystems is decisive for soil and vegetation activity to also be seasonal [42] as growth respiration is mainly affected by temperature [8]. Thus, in spring, it is expected that the highest stem respiration will be obtained, as it is the season of greatest stem diameter increments [6] and growth activity [7]. On the contrary, our study reflected that in autumn, and especially in winter, the respiratory activity of junipers was slowed down. In general, respiration measurements taken in the middle of April and at the end of September can be considered to be in the growth period [3].
It is noteworthy that in summer, the levels of stem respiration were greatly reduced in mature woodland. These results are in accordance with Rodríguez-Calcerrada et al. [5], which indicated that stem CO 2 efflux decreases with drought in Mediterranean forests. Given the scarcity of water resources in this site with the worst soil, it is likely that older junipers adapt to episodes of water stress by reducing their respiration. In general, water stress consists of a decrease in growth, and therefore the growth component of respiration should decrease [5].
Nonetheless, our study has also shown that in summer, stem CO 2 efflux was important in the young woodland. In general, as occurred in our results, previous research on the effect of water stress on respiration frequently show contradictions: while studies have shown decreased respiration [5], others have shown unaffected rates or an increased rates [43]. For example, at more severe levels of water stress, maintenance respiration may increase as a consequence of osmoregulation [44]. However, species that show slow growth rates (such as older juniper trees) present minor variations in response to stress than faster-growing species [43], such as young junipers.
Another hypothesis to explain the maintenance of stem respiration in summer for the young woodland is related to the quality of the soil. We hypothesized that water availability seems to be essential to maintaining growth respiration. In the woodland with better soil, vegetal activity could be continued in summer due to a greater reserve of water and nutrients. Several authors have also shown that the growth respiration of stems continued for 4 or 6 weeks after diameter growth ceased [45]. In the study site, the soil bedrock is a fissured and karstified dolomite, which permits the formation of small water pockets, and juniper vegetation probably obtains water from the deeper soil profiles in the summer [26]. Results in similar Mediterranean environments confirmed that deep-rooted trees improve water availability in dry seasons [46]. In the dry season, high temperatures can accelerate the process of autotrophic respiration [2]. At the end of this season, the more-or-less constant decline in respiration begins and continues until winter [8].

Effects of Stem Temperature on Stem CO 2 Efflux
In the present study, the stem temperature accounted for 71% of the variations in stem respiration, in line with previous results found in semi-arid climate; for example in Pinus canariensis [6] and in Pinus halepensis [40]. Stem temperature is closely related to and is an efficient predictor of stem respiration, better than other environmental variables such as soil water content [7], in accordance with our results.
The two values of Q 10 obtained for each juniper woodland were within the range of 1.0-4 reported for other conifer species [6,32,47]. For most species and organs, the Q 10 factor varies between 1.9 and 2.8 when respiration is measured between 5 and 30 • C [9], similar to the range in our study. However, in both juniper woodlands, and especially for the older forests, the Q 10 temperature coefficients are low (<2), which is the value usually taken as the reference mean [9,48].This is in accordance with the low values of stem CO 2 efflux measured. The respiration rate obtained in our research was more typical of cold ecosystems, i.e., forests with low productivity [9,47]. Juniperus has its origin in the Tertiary where it grew under colder climates [19], and slow-growing species show slower respiration rates even with high water availability [49]. The environmental limitations resulting from climatic extremes, such as very cold winter and summer drought, typical of semi-arid climates, condition the respiration-temperature relationship [6]. The potential phytoclimatic of this species is related to severe continental climates [22], such as those predicted in the climate change scenario [23]. Q 10 quotient may vary depending on the season, growth rate, or acclimatization situations [50]. Based on our results, the coefficient Q 10 in juniper woodlands depends on the site. Thus, our results have concluded that respiration rates vary within the same species in function of contrasting ecological requirements (soil quality), as has been previously cited [16]. Finally, our data showed that respiration at 0 • C was significantly higher in the young woodland (0.70 vs. 0.62 µmol CO 2 m −2 s −1 ). In accordance with Zha et al. [7], we can assume that respiration in this non-growing period (stem temperature of 0 • C) can be correlated with the maintenance respiration. This reflected that trees with higher growth rates exhibit higher maintenance respiration than slower growing ones, suggesting a balance between C supply and demand mediated by growth [16]. The other fraction of measured stem CO 2 efflux reflects the growth respiration, the energetic cost of constructing the new tissues [38], which is higher in the young woodland.

Conclusions
Our results confirm the hypothesis that in the juniper stands there is a high correlation between stem respiration and forest productivity. Thus, stem CO 2 efflux measured at diameter at breast height is an estimator of forest productivity in Spanish juniper woodlands. Therefore, our study represents an advance in knowledge about the variation in stem respiration in classic juniper forests (mature and young woodlands) growing at their distribution limit in a semi-arid climate. Stem temperature and seasonality are decisive factor to understand the variations in stem respiration of this juniper species.
The uniqueness of this juniper species has again become apparent in our study. Their low respiratory rates (especially in the older stands) and temperature coefficients are typical of species linked to cold climates, and scarcely productive forests. The slow-growing behavior of this species has been explained in terms of stem respiration. This is the result of its adaptability to a severe climate, such as the semi-arid climate of Southern Spain (hot, dry summers, and very cold winters with low rainfall). Thus, we think that this species could be utilized for forest restoration in sites with low potential productivity, such as in arid and semi-arid areas, and in the context of climate change.